Skip to content

Add kp-detect-v15 submission#155

Open
kprofundis wants to merge 3 commits into
liamdugan:mainfrom
kprofundis:kp-detect-v15-submission
Open

Add kp-detect-v15 submission#155
kprofundis wants to merge 3 commits into
liamdugan:mainfrom
kprofundis:kp-detect-v15-submission

Conversation

@kprofundis
Copy link
Copy Markdown

Adds kp-detect-v15: per-(domain x generator) base scorer trained on 1M balanced text rows + calibration. predictions.json covers all 672K test ids (id + score only). metadata.json follows three-field minimum. Author: Kareem Elsamadicy (Independent Researcher) — kelsamadicy@gmail.com.

@github-actions
Copy link
Copy Markdown

ghost commented May 23, 2026

Eval run succeeded! Link to run: link

Here are the results of the submission(s):

Kareem Elsamadicy (Independent Researcher)

Release date: 2026-05-23

I've committed detailed results of this detector's performance on the test set to this PR.

On the RAID dataset as a whole (aggregated across all generation models, domains, decoding strategies, repetition penalties, and adversarial attacks), it achieved an AUROC of 96.44 and a TPR of 89.32% at FPR=5% and 80.87% at FPR=1%.
Without adversarial attacks, it achieved AUROC of 97.46 and a TPR of 92.31% at FPR=5% and 84.80% at FPR=1%.

If all looks well, a maintainer will come by soon to merge this PR and your entry/entries will appear on the leaderboard. If you need to make any changes, feel free to push new commits to this PR. Thanks for submitting to RAID!

@kprofundis
Copy link
Copy Markdown
Author

can we merge this?

@kprofundis
Copy link
Copy Markdown
Author

Hi RAID maintainers — if the eval results above look good (AUROC 96.44 with adversarial / 97.46 without; TPR@5% 89.32 with / 92.31 without; TPR@1% 80.87 with / 84.80 without), I'd appreciate a merge so this baseline is locked in on the leaderboard before I push the v18 successor PR later today.

Thanks!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant